mosaicml/llm-foundry

Pull Requests Commits
Adding Simplified Coding Tasks (#645)
Michael Carbin committed 2 years ago
Verified cdb1c284
Add test suite for flash attention 2 (#666)
dakinggg committed 2 years ago
Verified 0045ae60
Inverse Square Root LR Schedule (#657)
Mansheej Paul committed 2 years ago
Verified 6c982767
put target back (#668)
dakinggg committed 2 years ago
Verified bdac4c71
fix (#667)
dakinggg committed 2 years ago
Verified a1283403
Add images with flash attention 2 (#651)
dakinggg committed 2 years ago
Verified ba6b8803
chang bsz to 1 (#654)
bmosaicml committed 2 years ago
Verified 1bf0a939
Switch ordering of foundry gpu tests (#665)
j316chuck committed 2 years ago
Verified b0bc6640
Revert "Clean up processes between distributed gpu tests (#660)" (#662)
j316chuck committed 2 years ago
Verified 11e26c18
Clean up processes between distributed gpu tests (#660)
j316chuck committed 2 years ago
Verified 03744477
Fix lion8b error correction with torch 2.1 (#656)
dblalock committed 2 years ago
Verified aa2ba9f1
Only log "changing autoresume" when actually changing (#653)
aspfohl committed 2 years ago
Verified df945fa2
Fixes to lion8b test for torch 2.1 (#649)
dakinggg committed 2 years ago
Verified 7fb084a4
add 2.1 images (#648)
dakinggg committed 2 years ago
Verified d3c33059
Fix overriding of rope_scaling config (#644)
dakinggg committed 2 years ago
Verified cb1d94ae
fix pins (#646)
dakinggg committed 2 years ago
Verified cf015dd6
Add flag to disable train metrics (#642)
mvpatel2000 committed 2 years ago
Verified 9025b833
Set pytest tmp path retention policy to None
j316chuck committed 2 years ago
Verified a0e64ba5
Merge release 0.3.0 back to main (#635)
dakinggg committed 2 years ago
Verified 7f7c0979
Multi eval dataset logging (#603)
snarayan21 committed 2 years ago
Verified 3d4fa0f4
Add code eval (#587)
samhavens committed 2 years ago
Verified fd36398d
bump composer version (#630)
dakinggg committed 2 years ago
Verified 7a8c1a51
Add node rank to signal paths (#629)
mvpatel2000 committed 2 years ago
Verified 36b8ccbf
Change repeat to expand in GQA (#628)
sashaDoubov committed 2 years ago
Verified 61dfbd6e
Propagate bias through model (#627)
mvpatel2000 committed 2 years ago
Verified 547b3137
Replace dashes with underscores in split name (#626)
irenedea committed 2 years ago
Verified f488ad51
Update setup.py to bump flash attn (#615)
mvpatel2000 committed 2 years ago
Verified 68835627
Fixes a bad merge in the tiktoken PR (#619)
dakinggg committed 2 years ago
Verified 44028316
Support for using tiktoken tokenizers (#610)
dakinggg committed 2 years ago
Verified a8e87835
no deps (#617)
dakinggg committed 2 years ago
Verified 512d4eaf
Terms Of Use Privacy Policy Refund Policy Impressum